The experiment randomly assigned people to groups and then increased the cognitive load on one group. Then it found a statistically significant difference between groups in a particular measurement.
This sentence is true. So as cognitive load increased, there was a significant difference in the choice of the two groups. What it doesn't seem to justify is whether or not it was related to willpower.
We also don't know how many chose cake in group A (2-digit group). If it was two in group A and 3 in group B (the 7-digit group), then there's your 50% increase.
Now as to your main point, what is willpower other than the applied ability to make good choices in the face of temptation? It's true that if you take this experiment in isolation you can only reasonably extrapolate to food. But that's where you bring in other studies, like the puzzle-solving dogs. The way you could disprove this theory is by doing a few similar tests with other willpower-related choices. Right now the evidence points in a certain way, but it's completely falsifiable and foobarbazqux's argument is an unrelated objection pertaining to where willpower comes from. foobarbazqux is the the one bringing the wishy-washiness into this argument that they blame psychology for.
As for the puzzle-solving dog study, I still don't think that implies a willpower/cognitive task difficulty link. It certainly doesn't help prove human decision making. I think we can both assume that humans are slightly more complex.
Finally, I also think linking this to app design is silly at best. But that is completely biased, personal opinion.